【代码贡献】完善 QARM 高阶关联规则枚举

[woahwhattheheck]

Summary

Related to #13.

QARM currently derives confidence rules by comparing each frequent k-itemset only with frequent (k-1)-itemsets. For a frequent 3-itemset such as {A,B,C}, that emits the 2→1 rules but misses valid 1→2 rules such as A->B,C.

This contribution keeps the existing quantum frequent-itemset search unchanged and extends only association-rule enumeration:

• enumerate every non-empty proper subset of each frequent itemset as a possible antecedent;
• reuse the already-computed frequent-itemset support table for confidence;
• preserve the existing two-decimal confidence semantics and public dict result shape;
• sort antecedent/consequent item ids when formatting rule keys so output is deterministic;
• expose the enhanced implementation through the existing pyqpanda_alg.QARM.QuantumAssociationRulesMining package API while retaining the original QARM implementation as the base class.

Validation

Focused deterministic fixture validation covers all 12 expected rules across frequent 2- and 3-itemsets, higher-order rules under a confidence threshold, and deterministic key formatting for unordered sets. Syntax compilation also passes. The current local runtime does not include pyqpanda3, so I am not claiming a full native repository test run; fork-hosted validation is queued and maintainer CI remains authoritative.

This scope is intentionally separate from #50: that PR changes QARM search correctness in qarm.py and explicitly leaves higher-order _get_all_conf() rule enumeration unresolved. The paths are disjoint, and this subclass composes on top of the base QARM implementation.

Base develop: 5f973efccb84bc193157d1ccebe32137e307293b. Exact head: c74d9273bdd87d4e9aacacceb25cf155c9d8dbc8. Diff: 3 paths, +111/-1.

Prepared with AI assistance for TokenJunkieLabs; operated by @woahwhattheheck.